search

microsoft / WSL

Issues found on WSL
https://docs.microsoft.com/windows/wsl
MIT License
17.3k stars 814 forks source link

WSL2 HTTPS Connections Timeout #6416

Open edarioq opened 3 years ago

edarioq commented 3 years ago

Environment

Microsoft Windows [Version 10.0.19042.685]
Release: 20.04
WSL2
Linux version 4.19.128-microsoft-standard (oe-user@oe-host) (gcc version 8.2.0 (GCC)) #1 SMP Tue Jun 23 12:58:10 UTC 2020

Steps to reproduce

For example, trying to install NVM fails with a timeout error.

curl -o- https://raw.githubusercontent.com/nvm-sh/nvm/v0.37.2/install.sh | bash

After further debugging I found that the script just tries to clone the repo using https. I then tried to clone any repo via https and it fails with a timeout error, SSH works perfectly fine.

WSL logs: https://gist.github.com/edarioq/0198119f6c461b4025a3344a354b8d20

Expected behavior

https connections should work and git clone should work without any timeouts.

Actual behavior

fatal: unable to access 'https://github.com/nvm-sh/nvm.git/': Operation timed out after 300038 milliseconds with 0 out of 0 bytes received

Things I've tried:

edarioq commented 3 years ago

Awesome support from the Microsoft team, guess it's back to good old Mac OS for my development needs.

OneBlue commented 3 years ago

Thanks for reporting the issue @edarioq .

Can you please follow these instructions and share the script output, and both wsl.etl and packets.etl ?

FremyCompany commented 3 years ago

Hi @OneBlue,

I'm a former Microsoft employee. I filed feedback about what might be a similar issue here: https://aka.ms/AAaztwh (and I included networking diagnostics).

Let me know if you need any help looking at this, I am willing to run further diagnosis if needed, as this is a blocking issue for me at the moment.

While my computer is in the hung network state, it is probably spamming the network one way or another, because my mom can't watch the television until I unplug the network cable and replug it.

FremyCompany commented 3 years ago

In case anyone wonders what to do in this situation, my solution for now has been to install Telerik Fiddler on the Windows side, and setup my Linux machine to use it as a proxy. Because of that trick, all traffic to the web is routed through Windows instead of Linux, and that has prevented the network connection from hanging so far.

Additional instructions
1. Install Telerik Fiddler 2. Configure Fiddler to enable proxying VMs 2.1. Enable external connections to the proxy 2.2. Enable interception of HTTPS traffic 2.3. Enable response streaming (New Fiddler version: click on the first button of the toolbar (called "Stream responses" on hover)) 3. Export Fiddler's certificate to your desktop, then install that certificate as root on Linux ``` sudo mkdir /usr/share/ca-certificates/extra sudo cp /mnt/c/Users/franc/Desktop/FiddlerRootCertificate.crt /usr/share/ca-certificates/extra sudo dpkg-reconfigure ca-certificates ``` 4. Enable connections to go through your firewall ``` In the Windows firewall, add an inbound rule to enable port 8866 ``` 4. Find your IP address ``` grep -m 1 nameserver /etc/resolv.conf | awk '{print $2}' ``` 5. Enable the proxy temporarily for the system: ``` export http_proxy=172.25.112.1:8866 export https_proxy=172.25.112.1:8866 ``` 6. Enable the proxy for apt-get ``` sudo nano /etc/apt/apt.conf ``` ``` Acquire::http::Proxy "http://172.25.112.1:8866"; Acquire::https::Proxy "http://172.25.112.1:8866"; ``` 7. Enable the proxy for pip ``` mkdir /home/fremy/.config/pip/ vi /home/fremy/.config/pip/pip.conf ``` ``` [global] trusted-host = pypi.python.org pypi.org files.pythonhosted.org ```
ppejovic commented 3 years ago

I've been experiencing something potentially similar for the past 2-3 weeks (I can't see the Feedback Hub submission @FremyCompany) . Larger downloads with curl will slowly degrade in download speed until the speed reaches zero, then curl just hangs without the download ever completing. It's possible this is affecting all downloads, it's just that the behaviour is easily observable with 30+ MB downloads. This has completely blocked me from using WSL, however @FremyCompany's fiddler proxy workaround has actually made it useable.

However, in my case I don't believe this is WSL specific as I've observed the same behaviour in an Ubuntu Hyper-V VM. Since WSL2 uses Hyper-V and generates a dedicated WSL Hyper-V switch I can only assume it is something in the Hyper-V virtual network stack.

Here are a couple of downloads that typically stall:

curl -L https://awscli.amazonaws.com/awscli-exe-linux-x86_64.zip -o /dev/null curl -L https://github.com/backstage/backstage/archive/master.zip -o /dev/null

I have tried multiple things to address this (recreating the virtual switch, disabling AV), however I have experienced this on multiple machines including a freshly re-imaged Windows 10 20H2. I'm in a corporate environment so the desktops get installed with an array of bloatware but at this point the only culprit is a Jan 2021 Windows Patch that mentions security updates to "Windows Virtualization": https://support.microsoft.com/en-us/topic/january-12-2021-kb4598242-os-builds-19041-746-and-19042-746-ab18a1a1-d572-598f-4d86-7137aad34056

dlakatos847 commented 3 years ago

I think the issue is more generic than some HTTPS URLs. I suspect it's a Hyper-V related TCP NAT problem. ICMP echo requests are transmitted and responded and UDP-based DNS queries can succeed also. No TCP connection can be established though (eg. with nc). I'm on Windows 10 x64 Pro 20H2 19042.928, WSL2 running OpenSuSE Tumbleweed.

dlakatos847 commented 3 years ago

Windows update resolved my TCP NAT problem. I'm on Windows 10 x64 Pro 20H2 19042.964 now.

FremyCompany commented 3 years ago

I think this issue can be closed indeed. There was a networking driver update and that fixed this issue for me as well, quite a while ago in fact.

ecalcutin commented 3 years ago

Might it be related to https://github.com/microsoft/WSL/issues/7254 ?

rstone commented 2 years ago

Was related to an open VPN connection for me. Once that was closed no more timeouts.

bhargavmodak commented 2 years ago

Well I am getting the same issue; about an year later than the Original Post. I'm on Microsoft Windows 11 Pro | Version 10.0.22000.493, and I run into the same issue on WSL2.

I think this issue can be closed indeed. There was a networking driver update and that fixed this issue for me as well, quite a while ago in fact.

An automatic one by Windows update, or a manual one?

Update :

I went to the actual Github repository and there they had this option : wget -qO- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.1/install.sh | bash

It seemed to work, and $ command -v nvm returns nvm after a restart of the terminal. I do not know if this is the right way to do this, however.

kmshort commented 2 years ago

I'm having this exact same problem now. Fresh Windows 11 Enterprise update, running WSL2/g. wget downloads sometimes time out, sometimes they works In other software, R, the same thing happens when grabbing packages from online repositories. It's a big problem.

bhargavmodak commented 2 years ago

Update :

I went to the actual Github repository and there they had this option : wget -qO- https://raw.githubusercontent.com/nvm-sh/nvm/v0.39.1/install.sh | bash

It seemed to work, and $ command -v nvm returns nvm after a restart of the terminal. I do not know if this is the right way to do this, however.

It is not. Any future installations of node suffer timeouts. It was strange to have my git on WSL and Node on Windows, so I had to reinstall git for windows. While WSL is good for simulating other features in Linux, not having nvm and Node.js really restricts any development environment.

brunoAltinet commented 2 years ago

If someone stumbles here, i couldnt get brew package to install, so fiddling with MTU seems to fix it (althought i did try a couple of things). https://github.com/microsoft/WSL/issues/4698#issuecomment-814259640 Note that on my side mtu 1350 vEthernet (WSL)

znmeb commented 2 years ago

I've got this now, I'm not sure when it started; I've only been doing WSL-intensive networking for a week or so. System specs:

Edition Windows 11 Pro
Version 21H2
Installed on    ‎2022-‎01-‎15
OS build    22000.613
Experience  Windows Feature Experience Pack 1000.22000.613.0

This is WSL preview - the one that supports GUI applications. I haven't tried dropping back to the default version. Here's the diagnostic run (network.log is the logfile of the network.bat run).

diagnostics.zip

plabon-asad commented 2 years ago

WSL2-network-connection-timeout

I am getting this timeout issue. Any solution guys?

CoryKniefel commented 2 years ago

If anyone end up here, make sure you try updating your kernel (powershell as admin): wsl --update.

znmeb commented 2 years ago

If anyone end up here, make sure you try updating your kernel (powershell as admin): wsl --update.

  1. Doesn't that happen automatically if you select "update all Microsoft products" in your update settings?
  2. Is this bug still open? It's been a while since I did any intensive downloading in WSL.
CoryKniefel commented 2 years ago

If anyone end up here, make sure you try updating your kernel (powershell as admin): wsl --update.

  1. Doesn't that happen automatically if you select "update all Microsoft products" in your update settings?
  2. Is this bug still open? It's been a while since I did any intensive downloading in WSL.

Not sure how all that works. But wsl --status told me I had auto updates off, and that there was an update available. I've already ran all of the windows updates manually, and had the exact problem described by original post.

lemonsaurus commented 1 year ago

If anyone end up here, make sure you try updating your kernel (powershell as admin): wsl --update.

I was having this problem on a fresh install of Windows 11, with a fresh Ubuntu 22.04 distro, with all windows updates installed. wsl --update solved the problem for me. Thanks @CoryKniefel!

toadharvard commented 1 year ago

Same problem. wsl --update can't solve problem because The most recent version of Windows Subsystem for Linux is already installed.

root@laptop:~# curl -vfsSL https://apt.releases.hashicorp.com/gpg
*   Trying 65.9.44.100:443...
* Connected to apt.releases.hashicorp.com (65.9.44.100) port 443 (#0)
* ALPN: offers h2
* ALPN: offers http/1.1
*  CAfile: /etc/ssl/certs/ca-certificates.crt
*  CApath: /etc/ssl/certs
* TLSv1.0 (OUT), TLS header, Certificate Status (22):
* TLSv1.3 (OUT), TLS handshake, Client hello (1):
* TLSv1.2 (IN), TLS header, Certificate Status (22):
* TLSv1.3 (IN), TLS handshake, Server hello (2):
* TLSv1.2 (IN), TLS header, Finished (20):
* TLSv1.2 (IN), TLS header, Supplemental data (23):
* TLSv1.3 (IN), TLS handshake, Encrypted Extensions (8):
* TLSv1.2 (IN), TLS header, Supplemental data (23):
* SSL connection timeout
* Closing connection 0
curl: (28) SSL connection timeout

Any suggestions?

ralphv commented 1 year ago

Same issue, clean fresh machine, Windows 11... In my case when I do npm install, I get a lot of timeouts and blocking...

JoshTheDeveloperr commented 1 year ago

I'm not sure why but this Stack Overflow answer fixed the problem for me. https://stackoverflow.com/a/70782306

For some reason the name resolution for https://registry.npmjs.org isn't resolving or is just extremely slow when using the stock resolv.conf configuration.

znmeb commented 1 year ago

I'm not sure why but this Stack Overflow answer fixed the problem for me. https://stackoverflow.com/a/70782306

For some reason the name resolution for https://registry.npmjs.org isn't resolving or is just extremely slow when using the stock resolv.conf configuration.

Interesting ... why should changing the DNS server fix this? Is there something wrong with the automatically generated DNS server that WSL wrote into the file?

RingoDev commented 1 year ago

Also experiencing a similar problem:

EDIT: setting nameserver to 8.8.8.8 in resolv.conf resolved the issue temporarily

znmeb commented 1 year ago

Also experiencing a similar problem:

* pnpm installs show many `ERR_SOCKET_TIMEOUT` and unusually long install times.

EDIT: setting nameserver to 8.8.8.8 in resolv.conf resolved the issue temporarily

I still don't get why hard-coding the nameserver to Google's DNS is faster than using the internal nameserver that Windows / WSL provide. Where is the extra time going in the software as shipped? Should Microsoft change something?

Tabares commented 1 year ago

Please try the following in your WSL

sudo rm /etc/resolv.conf
sudo bash -c 'echo "nameserver 8.8.8.8" > /etc/resolv.conf'
sudo bash -c 'echo "[network]" > /etc/wsl.conf'
sudo bash -c 'echo "generateResolvConf = false" >> /etc/wsl.conf'
sudo chattr +i /etc/resolv.conf

I can install with curl.

Larenatos commented 1 year ago

Please try the following in your WSL

sudo rm /etc/resolv.conf
sudo bash -c 'echo "nameserver 8.8.8.8" > /etc/resolv.conf'
sudo bash -c 'echo "[network]" > /etc/wsl.conf'
sudo bash -c 'echo "generateResolvConf = false" >> /etc/wsl.conf'
sudo chattr +i /etc/resolv.conf

I can install with curl.

Using Google's DNS didn't resolve the issue for me. Still timing out despite working on windows.

➜  ~ curl -v https://raw.githubusercontent.com/ohmyzsh/ohmyzsh/master/tools/install.sh
*   Trying 185.199.108.133:443...
* TCP_NODELAY set
* Connected to raw.githubusercontent.com (185.199.108.133) port 443 (#0)
* ALPN, offering h2
* ALPN, offering http/1.1
* successfully set certificate verify locations:
*   CAfile: /etc/ssl/certs/ca-certificates.crt
  CApath: /etc/ssl/certs
* TLSv1.3 (OUT), TLS handshake, Client hello (1):
* Operation timed out after 300164 milliseconds with 0 out of 0 bytes received
* Closing connection 0
curl: (28) Operation timed out after 300164 milliseconds with 0 out of 0 bytes received
Driftminder commented 1 year ago

Hello,

I've got the same trouble and try all this solution:

Current : Windows 10 22H2 19045.2251

Everything doesn't work, at least i've found a solution for me, change the MTU of the eth0 in WSL to 1350, after this my apt update / upgrade / install work properly.

This is command to use

sudo ip link set dev eth0 mtu 1350

cgagnonqc commented 1 year ago

I have the same issue and tried all proposed solutions above. None worked. I also tried solutions there , none worked.

I end up trying removing Docker desktop for Windows and only installing it like a native Linux. I also activated systemd support. https://nickjanetakis.com/blog/install-docker-in-wsl-2-without-docker-desktop

I works well and didn't get any troubles.

lxe commented 1 year ago

I'm running into https timeouts on WSL with windows 11. Just checking in.

gyohza commented 1 year ago

I was having the same problem, but only after upgrading WSL to WSL2.

I tried changing SSH keys, switching to HTTPS, deleting the repo and failing to clone back again, blaming my VPN, punching my screen... the error wouldn't budge.

Then I just used powershell to downgrade it back:

wsl --set-version Ubuntu-20.04 1

And then everything started working again 🤡

portwatcher commented 1 year ago

sudo hwclock -s syncing system time with windows works for me

nicejin74 commented 11 months ago

Checking Docker desktop on

DominikM commented 8 months ago

Want to chime in here to add my own experience with this issue. All https requests would fail by timing out, and every fix listed here didn't help. It clearly wasn't a DNS issue because urls could be correctly resolved to IP addresses. Ultimately it turned out that that the Hyper-V firewall was blocking WSL in a way that caused it to fail. Running these commands in powershell fixed my issue:

Set-NetFirewallHyperVVMSetting -Name '{40E0AC32-46A5-438A-A0B2-2B479E8F2E90}' -DefaultInboundAction Allow
Set-NetFirewallHyperVVMSetting -Name '{40E0AC32-46A5-438A-A0B2-2B479E8F2E90}' -AllowHostPolicyMerge False

The first command modifies the firewall to allow all inbound connections by default. This should probably be a bit more precise to avoid unwanted traffic, but as a first pass it worked. The second command is necessary to prevent the host firewall policy from overriding the locally set values. After this, everything worked as expected.

fooyun commented 7 months ago

same problem, and still have no idea

CaoKha commented 7 months ago

installing with curl, with ethernet connection, first try failed but second try after Ctrl+C worked, still have no idea....

jess-disguise commented 4 months ago

I started getting many different but related errors where I previously had no issues. Setup is WSL2 using Ubuntu 22.04 + Windows 11, building Docker images that contain RUN composer install --ignore-platform-reqs --optimize-autoloader :

Setting RUN COMPOSER_IPRESOLVE=4 composer install --ignore-platform-reqs --optimize-autoloader as described in the composer troubleshooting docs fixed the errors for me where many, many other solutions did not work.